Ages at Onset of Anxiety and Depressive Disorders in Parkinson’s Disease

The neurodegenerative process in Parkinson’s disease (PD) has been postulated to start in the olfactory bulb, the lower brainstem, and the enteric nervous system (1, 2). Based on the Braak’s staging system (1), the synucleinopathy slowly spreads within the brainstem reticular core, attacking systems involved with mood, sleep, and attention, and eventually reaching motor systems in the substantia nigra. In the preclinical PD phase, clinical signs and symptoms are absent (3). This is followed by the prodromal phase, which is characterized by nonmotor symptoms and subtle motor signs, consistent with Braak neuropathological stages 1–2 (3). Overt motor symptoms are noted in the clinical phase, corresponding to Braak stages 3–6, when nigral cell loss has reached at least a 40% level (1, 3). The PD prodrome lasts approximately 10 years, although constipation and REM-sleep behavior disorder (RBD) can be present up to 20 years before motor signs become apparent (3). Prodromal markers are still under investigation, with hyposmia/anosmia and RBD being most strongly associated with risk for PD (3, 4). Other proposed markers include excessive daytime sleepiness, orthostatic hypotension, urinary dysfunction, erectile dysfunction, depression, anxiety, and cognitive deficits (3, 4).

Depressive symptoms can precede PD motor manifestations in up to 50% of patients (4, 5), but the interval between depression onset and PD diagnosis has varied across studies. Leentjens et al. (6) reported an average interval of 10.1 years, with a range of 1 month to 36 years. Depression has been shown to increase risk for PD. Large epidemiological studies have revealed odds ratios of 1.2–3.2, with the highest odds ratio within 1 year of depression onset (7–10). Several studies found this association was no longer significant when depression preceded PD diagnosis by more than 5 years (8, 11). Other authors found a weaker yet still significant association in patients who had depression more than 15 years previously (7, 9). The association of anxiety with PD has been less thoroughly studied. Lin et al. (12) reported hazard ratios of 1.27–2.36, which correlated with the severity of anxiety symptoms. This association remained significant when anxiety preceded onset of motor symptoms by up to 20 years (11), although more recent studies have not replicated this finding (8). A prospective study conducted in male health care professionals suggested that phobic anxiety was linked to later development of PD (13).

Previous studies exploring the PD prodrome have been mostly population based, with various inclusion criteria for motor and psychiatric diagnoses (7–11). With few exceptions (12, 14, 15), prior studies did not rely on diagnoses made by psychiatrists. The present study explored the lifetime prevalence and onset ages of psychiatric disorders among patients with PD referred for psychiatric evaluation at a U.S. academic deep brain stimulation (DBS) center. To our knowledge, this is the first study to compare the exact onset ages of anxiety and depressive disorders established through psychiatric evaluation with PD diagnosis ages.

Methods

Setting

The University of California San Francisco Movement Disorders and Neuromodulation Center (MDNC) provides comprehensive evaluation and management of patients with movement disorders (predominantly PD), many of whom are referred for DBS. The interprofessional team includes a geriatric psychiatrist, whose role is to help manage patients with complex psychiatric comorbidities as well as provide perioperative psychiatric care for patients who undergo DBS surgery.

Participants and Data Collection

This study was approved by the University of California San Francisco Institutional Review Board and consisted of a chart review of all patients with PD evaluated by the MDNC psychiatrist from October 2015 to January 2018. Psychiatric diagnoses were established through 90–120 minute clinical interviews conducted by the psychiatrist, using DSM-5 criteria (16). Collateral history was obtained from family members and, if applicable, other mental health providers (therapist, psychiatrist) involved in the patients’ care, with patients’ permission. When necessary, additional clarifying questions were asked during follow-up visits in order to fully explore psychiatric symptoms and chronology. This approach allowed for a robust diagnostic impression to be formulated.

Onset ages of psychiatric disorders were established during the initial evaluation; when available, exact ages were listed. If patients could not recall the specific time when their psychiatric symptoms first occurred, the onset decade was recorded (e.g., 0–9 years old, 10–19 years old). The year of PD diagnosis was identified through chart review from neurologists’ notes and corroborated with patient self-report during interview. Age at PD diagnosis was calculated based on patient date of birth and year of diagnosis.

At MDNC, all patients referred for DBS surgery undergo extensive neuropsychological testing. Results of this assessment are routinely reviewed and inform the psychiatric evaluation. For this study, neurocognitive disorder diagnoses were established according to DSM-5 criteria, also taking into consideration neuropsychological testing findings.

Typical Ages at Onset

Typical ages at onset in the general population for the most common psychiatric conditions in our sample—major depressive disorder (MDD), generalized anxiety disorder (GAD), and panic disorder—were established based on several large U.S. population-based studies, taking into consideration the median (50th percentile band) ages. These studies included the Epidemiologic Catchment Area Study, which took place between 1980 and 1985 and surveyed more than 20,000 individuals from five sites across the United States (17); the National Comorbidity Survey Replication, a study of approximately 10,000 people conducted between 2001 and 2003 (18, 19); and the 2001–2002 National Epidemiologic Survey on Alcohol and Related Conditions Wave 1, which surveyed 43,903 individuals (20). When onset ages differed across studies of reference, the largest study took precedence. General population median onset ages were determined as follows: 29 years old for MDD (19, 21), 32 years old for GAD, (19, 22) and 24 years old for panic disorder (19, 23). This methodology was previously used and has been described in detail elsewhere (24).

Statistical Analysis

Frequencies were tallied for patient sex, psychiatric diagnoses, and surgical status (deep brain stimulation, pallidotomy, or no surgery). The percentage of patients whose anxiety or depressive disorders onset preceded PD diagnosis was calculated as a percentage of all patients with anxiety or depressive disorders. Means and standard deviations were calculated for patient age and PD diagnosis age. In a first analysis, the mean onset ages of anxiety and depressive disorders were calculated and compared with mean ages at PD diagnosis for the subgroups of patients with anxiety and depressive disorders, respectively, using a one-sample t test. In order to make use of all the available data, these two comparisons were then repeated, imputing the missing onset age data based on the onset decade in two ways. First, the middle value for the decade was inserted, and in a second analysis, a random age from within the decade was imputed. Exact onset ages of MDD, GAD, panic disorder, and social anxiety disorder were then compared with PD diagnosis ages in the subgroups of patients with these specific diagnoses. An additional analysis compared the mean differences between the onset ages of anxiety and depressive disorders and PD diagnosis ages in men versus women, using a standard t test.

For a second analysis, the mean onset ages of anxiety and depressive disorders were calculated and compared with mean ages at PD diagnosis for only the subgroups of patients whose anxiety and depressive symptoms predated PD diagnosis, using a one-sample t test. This analysis included only patients with onset of anxiety and depressive disorders prior to PD diagnosis for whom exact onset age values were available.

A third analysis examined the median onset ages of MDD, GAD, and panic disorder among patients in our sample relative to typical median onset ages for the same psychiatric conditions in the general population, using the sign test (25). Cohen’s d values were also calculated to explore effect sizes (26).

Results

Demographic and clinical characteristics of the 108 patients included in this study are summarized in Table 1. Most patients were able to place the onset of their anxiety or depressive symptoms within a certain decade of their life. Fifty-two patients (72.2%) among those who had anxiety disorders and 74 (78.7%) with depressive disorders indicated exact onset ages. Sixty-six (61.1%) patients had DBS, two (1.8%) had pallidotomies, and 40 (38%) had not had surgery.

Table 2 summarizes the onset ages of anxiety and depressive disorders and PD diagnosis ages for patients with anxiety and depressive disorders, first in the overall sample, then only for the subgroups of patients whose psychiatric symptoms preceded PD diagnosis. Mean onset ages for both anxiety and depressive disorders were significantly lower than mean PD diagnosis age. With regard to specific diagnoses, MDD (t=3.67, df=38, p=0.0009, Cohen’s d=0.58) and social anxiety disorder (t=7.33, df=7, p=0.0002, Cohen’s d=2.6) first occurred significantly earlier than PD diagnosis. Results were not statistically significant for GAD. In addition, there were no significant differences between men and women when the mean onset ages of anxiety disorders and depressive disorders were compared with the mean ages at PD diagnosis.

When restricting the analysis to patients whose anxiety and depressive disorders preceded PD diagnosis, age differences greatly increased. For patients whose anxiety preceded PD diagnosis, the mean onset age of anxiety disorders was 25.6 years lower than the mean PD diagnosis age (t=7.13, df=24, p<0.0001, Cohen’s d=1.4). For patients whose depressive symptoms predated PD diagnosis, the mean onset age of depressive disorders was 17.6 years earlier than the mean PD diagnosis age (t=7.60, df=30, p<0.0001, Cohen’s d=1.4).

Table 3 shows the mean differences between the onset ages of anxiety and depressive disorders and PD diagnosis ages for the respective subgroups and results of t test comparisons, first using exact onset ages as available, then using imputed age values. The mean onset age of anxiety disorders was 8.3 years lower than the mean PD diagnosis age in patients with anxiety disorders (t=2.91, df=51, p=0.0054). In patients with depression, the mean onset age of depressive disorders was 3.5 years earlier than the mean PD diagnosis age (t=2.02, df=73, p=0.0474). Using imputed values for onset ages, the mean differences increased to approximately 10 years for anxiety and 7 years for depression (all p values <0.0001).

Table 4 shows the median onset ages of MDD, GAD, and panic disorder in our study sample, compared with the median typical onset ages in the general population. Median onset ages in our sample were significantly higher than in the general population for all three conditions: MDD (median, 14.5, p<0.0001), GAD (median, 6.5, p=0.0002), and panic disorder (median=6, p=0.0005).

Discussion

To our knowledge, this is the first study to compare the exact onset ages of psychiatric disorders established through clinical interview with patient’s ages at PD diagnosis. Over half of patients in our sample experienced anxiety (55.5%) and depressive disorders (52.1%) prior to being diagnosed with PD. This is consistent with our previous observations in a subgroup of this sample (5), yet higher than reports from other authors, who noted depression preceding motor symptoms in only 30% of PD patients (4). In the present study, anxiety disorders manifested on average approximately 8 years before PD diagnosis, and depressive disorders predated PD diagnosis by about 3 years. When restricting the analysis to patients whose anxiety and depressive disorders preceded PD diagnosis, the interval between the onset of psychiatric diseases and PD diagnosis increased to approximately 25 years for anxiety disorders and 17 years for depressive disorders. By doing so, we excluded patients whose anxiety or depressive symptoms could have represented an adjustment to being diagnosed with a neurodegenerative disease or nonmotor fluctuations, such as wearing-off related anxiety. Our findings are consistent with the extant literature on PD depression, showing an interval of up to 36 years between the first depressive symptoms and PD diagnosis (6). Pontone et al. (14) reported a median age at onset of 44 years for anxiety disorders in a group of 63 patients with PD, whereas the mean age at onset for motor symptoms was 57 years. Our results closely replicate these previous findings with regard to the age at onset for all anxiety disorders and GAD, although in our sample panic disorder had a later onset (median age, 59 years) compared with the median age of 42 years reported by Pontone et al. (14)

Onset of psychiatric disorders after age 45 is uncommon and warrants a thorough medical workup. When comparing median onset ages of MDD, GAD, and panic disorder in our sample with typical median onset ages for these same conditions, the three disorders were noted to occur significantly later in our patients compared with the general population. This delayed onset may indicate a different phenotype of anxiety and depression in PD linked to early neurodegenerative changes in the brainstem reticular core in Braak stages 1–3, where serotonergic, noradrenergic, and cholinergic systems are perturbed and eventually damaged. Also, in some cases it is possible that certain limbic structures, such as the amygdala, are more sensitive to brainstem projection changes, and thus affective symptoms appear earlier than motor dysfunction (1). This hypothesis is supported by both anatomical and functional data (2, 27–30).

Iranzo et al. (28) reported the case of a 69-year-old man who presented with RBD and subsequently developed hyposmia, constipation, depression, and mild cognitive deficits but did not have parkinsonism. Neuropathological examination revealed Lewy bodies in his peripheral autonomic nervous system, olfactory bulb, medulla, pons, substantia nigra pars compacta (with an estimated cell loss of 20%−30%), nucleus basalis of Meynert, and amygdala (28). A resting-state functional MRI (fMRI) study identified connectivity dysfunction in the striato-thalamo-pallidal network in 17 patients with RBD and/or anosmia, the two strongest markers of prodromal PD (29). Rolinski et al. (30) also found fMRI connectivity abnormalities in the striato-thalamo-pallidal network of patients with PD and similar changes in patients with RBD, suggesting that these functional connectivity abnormalities are present in the PD prodromal phase.

Psychiatric prodromes have been described in other neurodegenerative movement disorders, such as dementia with Lewy bodies (31), Huntington’s disease (32, 33), and fragile X-associated tremor/ataxia syndrome (FXTAS) (24). In a group of 26 patients with FXTAS, median onset age of MDD was 49.5 years old, significantly higher than in the general population but lower than the median onset age of tremor (54 years old) and ataxia (55 years old) (24). This is consistent with our findings in patients with PD, where MDD, GAD, and panic disorder manifested significantly later relative to the general population but significantly earlier than motor symptoms, pointing to a continuum of neurodegenerative changes across time, brain regions, and functional networks.

Patients described in this study had a higher lifetime prevalence of anxiety and depressive disorders compared with the general PD population. Lifetime prevalence of anxiety disorders in patients with PD has been reported to be nearly 50% (14, 34) and that of depressive disorders, 66% (35). In our study, the lifetime prevalence of anxiety disorders was 66.7% and that of depressive disorders was 87%, both higher than previously reported. Almost two-thirds of patients in our sample had undergone surgical treatment; this generally indicates more advanced disease, which has also been correlated with higher severity of depressive symptoms (35). In addition, patients referred to the MDNC psychiatrist had more severe psychiatric presentations. As such, they may have been motivated to get better and reported their symptoms more openly. In general neurology or movement disorder practices, even though clinicians routinely inquire about nonmotor symptoms, patients may be reluctant to share their emotional struggles due to stigma (10).

With regard to prevalence of specific anxiety disorders, our findings partly overlap with those of Pontone et al. (14) The most common anxiety diagnoses in our sample were GAD (29.6%), panic disorder (15.7%), and social anxiety disorder (9.3%), while other anxiety disorders (including anxiety disorder due to another medical condition) reached 24.1%. Using DSM-IV-TR criteria, Pontone et al. found a similar social phobia lifetime prevalence (9%); however, the most common was anxiety disorder not otherwise specified (30%), followed by specific phobia (19%) and panic disorder (10%) (14). Longitudinal prospective studies in larger samples can further elucidate anxiety subphenotypes in patients with PD.

We did not find any gender differences with respect to the interval between the onset of anxiety and depressive disorders and PD diagnosis. PD is twice as common in men than women (36, 37), similar to the ratio found in our study (66.7% men) (Table 1). Men have also been shown to have an earlier PD onset and progress more rapidly (36). However, gender differences in the time lag between the first psychiatric symptoms and motor symptoms have not yet been elucidated. More research is necessary to clarify gender-specific nonmotor phenotypes in PD.

Our study had several notable strengths. The sample size of more than 100 patients allowed sufficient analytic power. Psychiatric diagnoses were established through clinical interviews, rendering a higher diagnostic accuracy compared with previous studies that did not include diagnoses made by psychiatrists, with several exceptions (12, 14, 15). Onset ages of psychiatric disorders were ascertained through psychiatric interview, aided by collateral history from family. Among prior studies, only one (14) relied on a psychiatrist to determine onset ages of anxiety disorders. Furthermore, to our knowledge the present study is the first to directly compare onset ages of anxiety and depressive disorders with PD diagnosis ages.

Our study also had limitations. This work took place in an academic DBS center; thus, the patient sample was not representative of the general PD population. A standardized psychiatric diagnostic instrument such as the Structured Interview for DSM-5 (SCID) (38) was not used. Because a single clinician collected all the data and individual interviewing styles vary, this could have led a type of cognitive bias, such as systematically asking or not asking about certain symptom clusters (39). The likelihood of such bias was mitigated by inviting family members to participate for parts of the interview, reviewing multiple additional sources of information (neurologists’ notes, neuropsychological evaluation, outside records), and communicating with other mental health providers involved in the patients’ care, as appropriate.

Only approximately 70% of patients in the present study indicated the exact ages when their psychiatric symptoms first occurred. This is not uncommon; in a study examining GAD in older adults, Chou (40) found that up to 9% of participants could not recall their ages at symptom onset. About 40% of patients in our sample had cognitive deficits, of whom 13% had major neurocognitive disorders (Table 1). This likely resulted in a recall bias. Cognitive deficits are prevalent among PD patients, with mild cognitive impairment (the closest diagnostic category in DSM-5 is mild neurocognitive disorder) affecting 40%−50% of patients and dementia (major neurocognitive disorder) occurring in up to 80% of patients over time (41, 42). As described above, multiple data sources were used in our study so as to limit the impact of patients’ cognitive deficits on psychiatric diagnostic accuracy. Another reason for missing onset age data could be the high acuity of psychiatric presentations. Several patients were seen as urgent consultations for suicidality, psychosis, or aggressive behavior. In these cases, the evaluation was focused on safety, which took precedence over obtaining a thorough symptom timeline.

Another limitation of our work is the observational design; a prospective study would have minimized recall bias and provided additional information. Most studies to date have been case-control or retrospective cohorts (10), with only two prospective longitudinal studies (13, 15). Lastly, we compared onset ages of psychiatric disorders with PD diagnosis ages, not the earliest age when patients experienced motor difficulties. PD is usually confirmed within 1 year of motor symptom onset (43); similar to previous studies (6, 8, 11), the data point most consistently available in the neurologists’ notes for our report was the year of PD diagnosis.

Conclusions

Anxiety and depressive disorders often precede PD motor symptoms but tend to occur later than in the general population. This suggests that neurodegenerative changes are present in parts of the brainstem reticular core and limbic system before motor circuits are affected to a degree that causes motor manifestations. Psychiatrists should keep in mind that onset of MDD, GAD, and panic disorder after age 45 might be a harbinger of a neurodegenerative movement disorder such as PD. Future prospective studies in larger samples are needed to clarify this particular phenotype of anxiety and depressive disorders and develop best evidence guidelines for managing them.